Methamphetamine Accelerates Cellular Senescence through Stimulation of De Novo Ceramide Biosynthesis

Methamphetamine is a highly addictive psychostimulant that causes profound damage to the brain and other body organs. Post mortem studies of human tissues have linked the use of this drug to diseases associated with aging, such as coronary atherosclerosis and pulmonary fibrosis, but the molecular mechanism underlying these findings remains unknown. Here we used functional lipidomics and transcriptomics experiments to study abnormalities in lipid metabolism in select regions of the brain and, to a greater extent, peripheral organs and tissues of rats that self-administered methamphetamine. Experiments in various cellular models (primary mouse fibroblasts and myotubes) allowed us to investigate the molecular mechanisms of systemic inflammation and cellular aging related to methamphetamine abuse. We report now that methamphetamine accelerates cellular senescence and activates transcription of genes involved in cell-cycle control and inflammation by stimulating production of the sphingolipid messenger ceramide. This pathogenic cascade is triggered by reactive oxygen species, likely generated through methamphetamine metabolism via cytochrome P450, and involves the recruitment of nuclear factor-κB (NF-κB) to induce expression of enzymes in the de novo pathway of ceramide biosynthesis. Inhibitors of NF-κB signaling and ceramide formation prevent methamphetamine-induced senescence and systemic inflammation in rats self-administering the drug, attenuating their health deterioration. The results suggest new therapeutic strategies to reduce the adverse consequences of methamphetamine abuse and improve effectiveness of abstinence treatments.


Introduction
The abuse of methamphetamine (D-meth) is a major health concern in industrialized countries, where a socially diverse group of users seeks the drug for its desirable psychological and physiological effects-a combination of euphoria, heightened arousal, reduced appetite and decreased fatigue [1]. Due to its highly addictive properties, D-meth also initiates an escalation in frequency and intensity of use, which brings about a host of negative symptoms such as panic and psychosis [1]. The pharmacological mechanism underlying these diverse actions is well understood-D-meth acts in the central nervous system to interrupt the reuptake of dopamine and other amine neurotransmitters, and facilitate their release into the synaptic space [1]. With prolonged drug exposure, these neurochemical alterations can lead to long-lasting damage to the brain, especially in structures containing dopaminergic axon terminals, which contribute to the emotional and cognitive problems experienced by D-meth addicts [2,3,4]. This transition to pathology has been attributed to a series of concurring events that include disruption of neuronal redox homeostasis [5,6,7,8], activation of apoptotic and necrotic processes [9], and recruitment of pro-inflammatory pathways dependent on the transcription nuclear factor NF-κB [7,10,11,12].
In addition to being neurotoxic, D-meth exerts widespread harmful effects throughout the body. Perhaps most striking among them is the extreme tooth decay ('meth mouth') that exacerbates the prematurely aged physical appearance typical of D-meth addicts [13]. While several abused drugs may increase the speed of organismal decline [14,15], post mortem studies have specifically linked the protracted use of D-meth to various pathologies characteristic of old age, including coronary artery atherosclerosis, pulmonary fibrosis and liver steatosis [16,17,18]. Nevertheless, the molecular mechanism through which D-meth might accelerate the emergence of age-related diseases remains unknown. Here we report that D-meth promotes cellular senescence and activates transcription of genes involved in inflammation and aging through a mechanism that requires increased biosynthesis of the sphingolipid messenger ceramide. These results shed new light on the molecular mechanism underlying D-meth toxicity and identify potential therapeutic targets to attenuate the life-threatening consequences of D-meth exposure in recovering addicts.

Surgery
Silastic intravenous catheters were implanted into the external jugular vein under anesthesia with a mixture of ketamine and xylazine (60 and 10 mg-kg -1 , i.p., respectively). Catheters exited the skin behind the ear. After catheter implantation, a nylon bolt glued to an acrylic mesh was implanted subcutaneously in the midscapular region. The nylon bolt served as a tether, preventing the catheter from being pulled out during self-administration sessions. Following surgery, the catheters were flushed daily during the first week with 0.2-0.3 ml of a solution containing cephalosporin (100 mg-ml -1 ; cefazolin USP; Hospira Inc., Lake Forest, IL, USA) and flushed with saline before and after each daily session to maintain patency.
Apparatus We used 18 standard operant-conditioning chambers (Coulbourn Instruments, Lehigh Valley, PA), which contained a white house light and two holes with nose-poke operanda on either side of a food trough. Upon activation, each nose poke produced a brief feedback tone. One hole was defined as active (left in 9 chambers, right in the remaining 9) and the other hole as inactive. D-meth or saline were delivered through Tygon tubing, protected by a metal spring and suspended through the ceiling of the experimental chamber from a single-channel fluid swivel. The tubing was attached to a syringe pump (Harvard Apparatus, South Natick, MA), which was programmed to deliver 2 s injections. The injected volume was adjusted for every animal to deliver a D-meth dose of 0.1 mg-kg -1 per injection. Experimental events were controlled by microcomputers using MED Associates interfaces and software (Med Associates Inc., East Fairfield, VT).
Procedure Each experimental group of rats was divided into 2 subgroups, which were tested at the same time. Subgroup 2 served as yoked control and passively received a saline injection each time a response-contingent injection of D-meth was self-administered by a Subgroup 1 rat. Nose-poke responses by yoked control rats were recorded, but had no programmed consequences. Eight consecutive 15 h sessions were always conducted between 4 pm and 8 am. At the start of each session, a white light was turned on and a priming injection of D-meth (0.1 mg-kg -1 ) or saline sufficient to fill the "dead" space of the catheter, was automatically delivered. Rats learned to self-administer D-meth under one-response, fixed ratio schedule (FR1). Each nose-poke response in the active hole (FR1) produced a delivery of D-meth injection (0.1 mg-kg -1 ) followed by a 30 s timeout, during which the chamber was dark and responses in either hole had no programmed consequences. Responses in the inactive hole were recorded, but had no programmed consequences.
Tissue collection The rats were killed by decapitation 2 h after the end of the eighth session. Brain, liver, heart, kidney (left), spleen, pancreas, testis, epididymal fat, skeletal muscle (vastus lateralis), and skin (hind paw) were harvested from each rat. The tissues were rinsed in a mix of RNasefree water with DEPC-treated PBS and blotted with sterile gauze. Brains and livers were snapfrozen in isopentane. All other tissues were snap-frozen in liquid nitrogen. Tissues were wrapped in aluminum foil and stored in -80°C.

Cell cultures
Immortalized mouse embryonic fibroblasts (MEF) were purchased from American Type Culture Collection (Manassas, VA), murine C2C12 cells were a kind gift of Dr. Maria Pennuto's group (Italian Institute of Technology, Genova, Italy), primary MEF cultures were prepared from C57BL/6 mouse embryos, as described [19]. See Supplemental information for detailed culture methods.

Lipid extractions and analysis
Lipid extractions were carried out as described [20]. A detailed description of the extraction procedure and LC/MS conditions used for lipid analysis is reported in Supporting Material and Methods.

Ceramide synthase activity
Ceramide synthase activity was measured as described [21]. A detailed description of assay conditions is reported in the Supporting Material and Methods.

Gene expression and silencing
See S1 Materials and Methods and S8 Table for detailed method.

Reactive Oxygen Species (ROS) Production
ROS production was measured using the fluorescent probe CM-H2DCFDA (Invitrogen). MEFs were grown in Dulbecco's Modified Eagle's Medium (DMEM) without phenol red. Cells were plated to subconfluence in 12-well plates, washed 3 times with pre-warmed Phosphate Buffered Saline (PBS) and loaded for 30 min at 37°C/5% CO 2 with 5 mM CM-H2DCFDA in DMEM without phenol red (loading medium). The loading medium was removed and prewarmed fresh medium containing the different cytochromes P450 (CYP) inhibitors in presence or absence of D-meth was added. Fluorescence (excitation at 485 nm, emission at 530 nm) was analyzed immediately, then cells were incubated at 37°C and 5% CO 2 , and fluorescence was measured at the indicated time points. ROS rate versus control (%) was calculated subtracting the percentage of ROS increasing from time zero in the D-meth-treated samples to the percentage of ROS increasing from time zero in the vehicle treated samples.

Chromatin immunoprecipitation (ChIP)
We used the two-step cross-linking method described in Nowak et al. [23]. See Supporting Material and Methods for detailed description.

Statistical analyses
See S1 Materials and Methods for details.

Methamphetamine self-administration increases tissue ceramide levels
To determine whether abnormalities in lipid metabolism might contribute to the harmful effects of D-meth, we conducted an unbiased lipidomic survey in organs and tissues of rats that selfadministered the drug intravenously [24]. A large percentage of animals taking D-meth escalated to high exposure levels over a few days (S1 Fig.), a behavior that closely mimics abuse of the drug in humans [24]. After 8 days of access to D-meth, the rats were euthanized and tissues were subjected to solvent extraction. The extracts were fractionated by liquid chromatography (LC) and representative lipid class constituents were identified and quantified by mass spectrometry (MS). The data were initially processed using heat-maps to facilitate visual inspection of broad regions of interest in the lipidome ( Fig. 1A-C). Subsequent quantitative analyses were conducted separately using appropriate internal standards ( Fig. 1D-G, S1-S4 Tables).

Methamphetamine stimulates de novo ceramide biosynthesis
We next examined whether acute exposure to D-meth might exert similar effects to those seen with D-meth self-administration. We used a non-contingent drug regimen involving two injections of D-meth separated by a 2 h interval, widely utilized to mimic the 'runs' wherein abusers binge on the drug [29]. When administered at a cumulative dose of 20 mg-kg -1 , D-meth increased ceramide content, CerS6 transcription and CerS activity in rat skeletal muscle (S2A-C Fig.). A lower cumulative drug dosage (10 mg-kg -1 in two injections of 5 mg-kg -1 each) increased CerS6 transcription (S2B Fig.) and produced a trend toward increased ceramide synthase activity (S2C Fig.), but had no statistically detectable effect on ceramide levels (S2A Fig.).
To further characterize the mechanism through which D-meth stimulates ceramide production, and eliminate possible confounders due to indirect effects exerted by the drug in vivo, we used cultures of mouse embryonic fibroblasts (MEF) and differentiated C2C12 myoblasts. Incubation of primary MEF with D-meth (1mM) elicited a rise in ceramide content ( Fig. 2A), which reached statistical significance 24h after addition of the drug (Fig. 2B). Similar results were obtained in an immortalized MEF line, where a detailed time-course analysis revealed that increases in ceramide levels were accompanied by parallel changes in the corresponding dihydroceramides, and were preceded by a rise in dihydrosphingosine (sphinganine), which both implicated de novo ceramide biosynthesis (S3 Fig.). Highlighting the selectivity of the effect of D-meth, two structurally related phenylethylamines, D-amphetamine and 4-hydroxy-D-methamphetamine, produced modest but significant changes in ceramide levels, while the enantiomer of D-meth, L-meth, was completely ineffective (Fig. 2C). The ability of D-meth to stimulate ceramide accumulation was confirmed using serum-differentiated C2C12 myoblasts. In these muscle-like cells, exposure to D-meth (0.1-1 mM) for 24 h caused concentrationdependent elevations in dihydrosphingosine, dihydroceramide (d18:0/16:0) and ceramide (d18:1/16:0), which were statistically significant at D-meth concentrations !0.5 mM (S4 Fig.).
In addition to increasing ceramide levels, in primary MEF cultures D-meth (1 mM, 24 h) also enhanced transcription of genes encoding for enzymes in de novo ceramide biosynthesis (Fig. 2D). Importantly, drugs that interrupt de novo ceramide production-including the SPT inhibitors, L-cycloserine and myriocin, and the CerS inhibitor, fumonisin B 1 [30,31,32]prevented D-meth-induced ceramide accumulation and decreased baseline ceramide formation ( Fig. 2E-G). In these experiments, D-meth (1 mM, up to 48 h) caused no overt membrane damage (Fig. 2H) or mitochondrial dysfunction (Fig. 2I) and no change in cell viability (Fig. 2J), indicating that its effects on ceramide were not secondary to cell injury or cell death. As expected, under the same conditions cisplatin, a classic chemotherapy drug, produced substantial cell death (S5 Fig.). Collectively, the results outlined above suggest that D-meth stimulates de novo ceramide biosynthesis by interacting with a structurally selective molecular target. Because of its presence in cells that are not neurons, such as fibroblasts and differentiated myoblasts, this putative target is likely to be distinct from the neuronal monoamine transporters responsible for the stimulant properties of D-meth [1].
Role of cytochrome P 450 and NF-κB D-meth is metabolized in humans and rodents by cytochrome P 450 (CYP)-2D6, a widely distributed CYP isoform that catalyzes the oxidation of D-meth into D-amphetamine and 4hydroxy-D-methamphetamine [33,34]. A by-product of this reaction is the formation of ROS [8], which are known to activate NF-κB-dependent stress-response signals that can lead to ceramide formation [35]. To determine whether CYP metabolism might be involved in D-methinduced ceramide production, we blocked CYP activity in primary MEF cultures using a panel of five chemically distinct agents: clotrimazole, SKF-525A and cimetidine (three pan-CYP inhibitors), quinidine (selective for CYP2D6) and HET-0016 (selective for CYP4A). LC/MS analyses showed that incubation of MEF in the presence of clotrimazole (1-10 μM) increased the levels of non-metabolized D-meth (Fig. 3A) and concurrently decreased D-meth-induced ceramide accumulation (Fig. 3B), while exerting no effect on baseline ceramide content (in pmolmg -1 protein, control: 52.9±5; clotrimazole 10 μM: 49.2±3; n = 3). Similarly to clotrimazole, SKF-525A, cimetidine and quinidine (each at 10 μM) normalized ceramide formation in D-meth-treated cells, whereas HET-0016 (10 μM) was ineffective (Fig. 3C).
As a further test of the role of ROS in D-meth-induced ceramide production, we determined whether exposure to D-meth stimulates ROS generation in MEF. As anticipated from previous studies [8], D-meth caused a concentration-dependent increase in ROS formation (Fig. 3D), whereas 4-hydroxy-D-methamphetamine (a product of D-meth metabolism via CYP2D6) [33,34] had little or no effect (Fig. 3E). The release of ROS evoked by D-meth was prevented by the pan-CYP inhibitors-clotrimazole (Fig. 3F), SKF-525A and cimetidine (Fig. 3G)-and the CYP2D6 inhibitor, quinidine (Fig. 3G), but not by the CYP4A inhibitor, HET-0016 (Fig. 3G).
An early cellular response to ROS formation is the recruitment of NF-κB [36,37], which can also be induced by D-meth [10,11]. Supporting a role for NF-κB activation in the effects of D-meth, we found that exposure of MEF cultures to the drug (1 mM, 24 h) resulted in a 4-fold increase in the amount of NF-κB p65 subunit bound to the TNF-α promoter (Fig. 3H). Moreover, pharmacological blockade of NF-κB using three different inhibitors-thalidomide (25 μM), 5-aminosalicylic acid (10 μM) and JSH-23 (10 μM)-prevented D-meth-induced ceramide accumulation (Fig. 3I-K) as well as D-meth-induced SPT and CerS transcription (S6 Fig.). Silencing expression of the p65 subunit of NF-κB with a selective short interfering (si) RNA (S7 Fig.) impaired the ability of D-meth (1 mM) to increase ceramide levels (S8 Fig.) and stimulate expression of enzymes of de novo ceramide biosynthesis (S9 Fig.). These findings are consistent with the possibility that CYP-mediated metabolism of D-meth induces formation of ROS and engages NF-κB, resulting in stimulation of ceramide biosynthesis.

Methamphetamine accelerates cellular senescence
A variety of stressful stimuli force proliferating cells into a state, called cellular senescence, characterized by cell-cycle arrest, apotosis resistance, and increased secretion of proinflammatory factors [38,39,40]. Since membrane-permeant ceramide analogs cause senescence in fibroblasts and other cell types [41,42,43], we asked whether D-meth-induced ceramide formation might exert a similar effect. For these experiments we used primary cultures of MEF, the senescence response of which has been most extensively characterized [41]. When cultured in control medium, primary MEF progressively developed a phenotype characterized by increased expression of senescence-associated β-galactosidase (β-Gal; Fig. 4A), flattened morphology (Fig. 4B), and higher levels of ceramide (Fig. 4C) and CerS5 mRNA (Fig. 4D). Culturing MEF in the presence of D-meth (1 mM, 48 h) intensified this senescent phenotype with decreased intensity as the cells approached the end of their proliferative capacity (Fig. 4A). The possibility that D-meth accelerates senescence, suggested by these observations, was further supported by results showing that treatment with D-meth reduced DNA synthesis (Fig. 4E), lowered the number of cell doublings (Fig. 4F), and stimulated the transcription of genes encoding for cell cycle-regulating proteins (p53, p21) and pro-inflammatory cytokines [interleukin-6 (IL-6) and tumor necrosis factor-α (TNF-α)] (Fig. 4G-J), which are all known to participate in senescence [38,39].
Next, to assess the contribution of ceramide to D-meth-induced senescence, we interrupted the de novo biosynthesis of this sphingolipid messenger using the SPT inhibitor, L-cycloserine (30 μM), or the CerS inhibitor, fumonisin B 1 (50 μM). The compounds suppressed senescenceassociated β-Gal expression in MEF treated with D-meth (1 mM, 48 h), but not in MEF treated with the cell-permeant ceramide analog C8 (10 μM, 48 h) (Fig. 4K-L). Additionally, L-cycloserine (30 μM) normalized transcription of p53, p21, IL-6 and TNF-α (Fig. 4G-J). The results suggest that D-meth accelerates cellular senescence in MEF through induction of de novo ceramide biosynthesis. This process appears to depend on the recruitment of NF-κB, as documented by the finding that thalidomide and siRNA silencing of NF-κB were each able to prevent it (S10 Fig.).

Ceramide is a key effector of systemic metamphetamine toxicity
The finding that D-meth promotes senescence in primary MEF cultures prompted us to ask whether self-administration of the drug might activate a transcription program suggestive of accelerated senescence in vivo [39]. Since many senescence-associated proteins are also involved in chronic inflammation [40], we reasoned that their abnormal expression might contribute to the systemic pathological state elicited by D-meth. As a first test of this idea, we quantified mRNA levels of various genetic markers of senescence and inflammation in rats self-administering D-meth. The transcription of six such markers-p53, p21, p16, IGF-1 [38,39,44]-was substantially increased in skeletal muscle after 8 days of voluntary D-meth intake, relative to yoked vehicle self-administration (Fig. 5A-F). Statistically significant elevations in TNF-α and IL-6 transcripts were also observed in heart and skin (S7 Table). These molecular changes were accompanied by a reduction in body weight (Fig. 5G), which is suggestive of sharp health deterioration. Acute non-contingent injections of D-meth, which closely mimic a drug binge (two 10 mg-kg -1 injections separated by a 2 h interval), increased transcription of p21 and IL-6 but had no effect on the levels of p53, p16, IGF-1 or TNF-α mRNA (Fig. 5H-M). This difference is noteworthy, because it suggests that prolonged intake of D-meth is necessary for the drug to fully engage molecular processes involved in senescence and inflammation.
Lastly, to evaluate the role of ceramide in systemic D-meth toxicity, we treated rats that selfadministered D-meth with the SPT inhibitor L-cycloserine (10 mg-kg -1 , twice daily for 2 days, followed by 20 mg-kg -1 twice daily for 3 days). The L-cycloserine regimen normalized ceramide content in skeletal muscle and brain tissue (dorsal striatum) of rats that self-administered D-meth, without changing baseline ceramide levels in animals that received yoked vehicle injections ( Fig. 6B-C, respectively). L-cycloserine did not alter key centrally mediated actions of D-meth-including its ability to maintain self-administration (Fig. 6A), increase body temperature (Fig. 6D) and reduce food intake (Fig. 6E). Nevertheless, the SPT inhibitor corrected the abnormalities in body weight (Fig. 5G) and gene expression caused by D-meth in both skeletal muscle (Fig. 6F-K) and dorsal striatum (Fig. 6L-O), providing evidence that ceramide is an obligatory effector of the systemic inflammation and health deterioration produced by D-meth.

Discussion
While conducting a survey of lipidomic abnormalities associated with self-administration of D-meth, we found that ceramide production is strongly increased in rats that voluntarily take the drug. This effect was clearly detectable in select regions of the brain, such as the dorsal striatum, but was particularly pronounced in peripheral organs known to undergo pathological changes following prolonged exposure to D-meth (e.g., skeletal muscle, heart and liver) [16,17,18]. In those organs, accrued ceramide biosynthesis was associated with activation of a senescence-like transcription program-characterized by enhanced expression of genes involved in cell-cycle control (e.g., p21, p16) and chronic inflammation (e.g., IL-6 and TNF-α)which could be recapitulated in vitro by treating mouse fibroblasts or differentiated C2C12 myoblasts with D-meth. Importantly, pharmacological inhibitors of ceramide formation prevented the occurrence of these molecular events, and lowered D-meth-induced cellular and systemic toxicity. These studies suggest that modulation of ceramide signaling might be utilized therapeutically to attenuate the serious, often fatal health consequences of D-meth abuse and enhance the success of behavior-based abstinence programs.
Our lipidomic analyses show that D-meth self-administration in rats-a model that realistically captures abuse of the drug in humans [24]-is accompanied by changes in the levels of ceramide and its precursor dihydroceramide, which are suggestive of an up-regulation of the de novo pathway of ceramide biosynthesis. This possibility is supported by two additional observations: (i) D-meth increased transcription of genes encoding for two key enzyme families in that pathway, SPT and CerS; and (ii) pharmacological interference with the activities of those enzymes blocked D-meth-induced ceramide formation. How may D-meth influence SPT and CerS expression? Pharmacological evidence implicates CYP-mediated metabolism, which is responsible for the metabolic clearance of D-meth in humans and rodents [33,34]. Experiments with mouse fibroblasts show that (i) D-meth stimulates ROS generation, NF-κB activation and ceramide production; (ii) genetic or pharmacological interference with CYP or NF-κB activities prevents these effects; (iii) structural analogs of D-meth, such as L-meth and 4hydroxy-D-methamphetamine, which are comparatively poor substrates for CYP-mediated oxidation, only partially mimic the effects of D-meth. Thus, an economical interpretation of the results outlined above is that D-meth metabolism via CYP generates ROS, which in turn activate SPT and CerS transcription through an NF-κB-dependent mechanism. This model is consistent with the greater vulnerability of human CYP2D6 extensive metabolizers to D-meth neurotoxicity [45], but does not exclude the possibility that D-meth might also stimulate ceramide biosynthesis through other mechanisms.
Ceramide plays an important role in the transition of cells into replicative senescence [26,28,46], a state in which mitotic cells irreversibly stop dividing and start secreting pro-inflammatory cytokines and tissue-modifying factors [39,40]. Our in vitro experiments indicate that the increased ceramide biosynthesis evoked by D-meth accelerates progression of a senescent phenotype characterized by lowered proliferation, altered cell morphology and heightened expression of β-Gal and other senescence-associated markers. Among such markers is a group of genes encoding for protein regulators of cell cycle (p53, p16, p21 and IGF-1) and inflammation (IL-6 and TNF-α), which we found to be abnormally high in skeletal muscle and other peripheral tissues of rats that voluntarily take D-meth. Importantly, this same panel of genes is known to be elevated in tissues of elderly persons [47,48,49] and is thought to participate in chronic inflammatory conditions that are typical of old age [40]. The premature occurrence of such conditions is a hallmark of D-meth abuse in humans, which cannot be explained by the psychostimulant and sympathomimetic properties of this drug [17,18], but is consistent with its ability to elevate ceramide levels. Indeed, overactive ceramide signaling has been implicated in several age-related pathologies-including coronary artery atherosclerosis, liver steatosis and pulmonary fibrosis [50,51,52]-which are frequently documented in autopsy reviews of deceased D-meth users [16,17,18]. Collectively, the available data suggest that the rapid health decline caused by D-meth is due to a ceramide-mediated acceleration of genetic programs that are also engaged during chronic inflammation and aging. Pharmacological strategies aimed at normalizing ceramide signaling-for example by modulating NF-κB recruitment (which include drugs that are already available in the clinic, such as 5-aminosalicylic acid and thalidomide) or SPT activity-might slow down this pathogenic process and facilitate detoxification of D-meth users.